Sunday 7 September 2008

Chess Ratings: A Necessary Evil?

Rating (noun): a classification according to order, rank or value ... an estimated value of a person's position (From Chambers 21st Century Dictionary).

Why do we have chess ratings? Shouldn't arguments over who is the best player be settled by direct competition across the board, not by a statistical calculation of probabilities (and I say that as a mathematics graduate).

I would not want to take anything away from Magnus Carlsen, but does the fact that his current live rating of 2791.6 is fractionally higher than Vishy Anand's of 2790.6 actually mean anything?

Arpad EloThe official "Elo" rating system, named after it's inventor, the Hungarian professor Arpad Elo (pictured), is a relatively recent phenomenon, only being officially adopted by FIDE in 1970. The chess world managed perfectly well without ratings before then, so why have them at all?

Of course, I'm partly playing devil's advocate here. Ratings are obviously useful to provide a benchmark to compare players' relative strengths, especially if they have never played one another before. But in a one-on-one adversarial contest like chess, surely ratings are wholly inadequate to describe the multitude of factors that come into play when two individuals with a variety of strengths and weaknesses face each other in combat?

It is my contention that chess ratings are being overused, misused and even sometimes abused, when they represent nothing more than a mathematical statement about players' past results. They assuredly do not prove that one player is better than another - only the estimated probability of the outcome of a game between them.

To contrast chess with another one-on-one contest, no-one cares if a boxer is ranked more highly than his adversary. It all comes down to what happens on the night. When standing toe-to-toe in the ring, 'rankings' count for nothing. So why are chess ratings given so much significance in comparison? In 1974, George Foreman would undoubtedly have been 'ranked' the best boxer in the world, but when Muhammed Ali floored Foreman in the eighth round of the Rumble In The Jungle, no-one questioned Ali's right to be known as the world champion.

Perhaps the answer is that being the 'world champion' and 'ranked the best in the world' are not always the same thing depending on the sport or game in question. Does it matter if the 'world champion' is not ranked number 1 in the world? If ratings matter so much in chess, why do we even have a World Chess Championship at all? Why not just declare the highest ranked player to be the world champion and save FIDE the expense of organising a world championship cycle?

This should never happen and with good reason. When Anand and Kramnik face each other in Bonn in October, they will be continuing a long chess tradition stretching back over a century with few interruptions, whereby a new champion must overthrow the old champion in a direct contest to prove his worth. Good luck to both players, and may the best (and not necessarily highest rated) man win!

4 comments:

chessx said...

In the days before ELO chess players knew who was good,very good,world champion contender not by rating points but playing strenght.
This was gained by tournament or match performances.
In todays chess elo ratings should only be used up to im title level.
After that an im or gm should be able to prove their worth by results.
Of course many would say elo is used to gain titles.
But Fischer,Tal and Spassky did not need elo points to get titles, they gained them through winning srong tournaments.
I once saw a list of past champions Capablanca and Alekhine had ratings of 2700 and 2750 does this mean Carlsen and some of the others players are better than them?
I know its different times but what is the use of comparasons if they dont mean much.
I have been wanting to say that for a long time,thanks.

Ryan said...

I think that it's been well documented that since the introduction of Elo ratings, there has been a noticeable inflation effect. There are currently 31 players over 2700 on the live rating list. This is crazy! They are all great players, but are they stronger than (say) Spassky at his peak? I certainly don't think so.

As for projecting ratings back in time to the days of Alekhine and Capablanca, I simply don't believe the figures are reliable enough to be meaningful. In reality, I don't think there is ANY way of comparing players from different eras. Ratings only have any merit to compare players of the same era against one another - and even then, the point I'm trying to make is that ratings are totally over-emphasised. What matters is what happens when two players clash over the board. I don't really much care who has the highest rating, but I'm utterly fascinated to see how Vishy and Vlad get on in Bonn!

chessx said...

That's what i wanted to say,it's just that you said it better than me,much better .
I am looking forward to the title match 2 players at on top of their game.
We should get some great games.

Ryan said...

Thanks Chessbumbus! You have a very nice blog too, so I've added you to my list of favourite blogs.

Cheers!